438 RAPID POPULATION GROWTH-I1
concluded that birth interval was not a significant factor. Yerushalmy, in discussing this (61), called attention to the time factor affecting prematurity in the brief interval group, as well as possible biases associated with admission policies which might affect a series of patients from only one hospital. He concluded that Eastman's findings did not necessarily contradict those of other investigators.
In the cross-sectional study of all births in England and Wales mentioned earlier, the effect of birth interval was also examined, in this case by another indirect method (27). Having data available concerning duration of marriage, the investigators assumed that the birth interval would be brief in the case of mothers giving birth to a second child after less than 2 years of marriage, or a third after less than 3 years. They therefore considered such births "closely spaced" and compared the mortality rates in that group with the rates in all other births. Their results are shown in Figure 12, where it may be seen that in all maternal age groups and in all social classes the postneonatal mortality rates are higher in the closely spaced group than in the other groups. These rates are also higher in younger mothers, as we saw in the previous data, and a third closely spaced child is at greater risk than a second such child.
They also found, it should be noted, somewhat higher rates among firstborns born after less than a year of marriage in all maternal age groups and all social classes than in first children born later. The differences here were not as great as those between "closely spaced" and "others," but they concluded that " 'duration of marriage', then, apparently reflects something in addition to birth spacing, since first children are also affected by it" (27, p. 105). Clearly children born early in the first year of marriage would either have been born prematurely and exposed to the risks of higher mortality associated with prematurity, or have been conceived premaritally. One can only speculate about the effects of premarital conception, but it seems likely that many mothers in such situations would receive less than optimal prenatal care and be subject to more emotional stress than mothers who conceive their first child after marriage.
Finally, Gordon and Wyon, in their prospective study of children born in Punjabi villages (18, 19) also sought an association between, birth interval and mortality rates. They compared the rates in their study children grouped according to the interval between the index child (the individual within a group under study) and the preceding sibling. The data are shown in Table 14. Both neonatal and infant mortality rates were appreciably higher among infants born at short intervals, and the differences observed, even in this relatively small population, were statistically significant.* It is interesting to